Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Helium Leak Testing for Hermetic Seals
Transcript
- Lucas: If these conversations are useful for what you're building or running, you're in the right place. Today we're talking about one of those tests that sounds like overkill until you've had a device fail in the field because moisture got inside an enclosure. Luna: Helium leak testing — I've seen it on spec sheets but never really understood when a startup should spring for it. Lucas: Right. So the basic idea: you want a hermetic seal — meaning absolutely no gas or liquid can get in or out. Think medical implants, outdoor sensors, underwater cameras. Even a tiny leak can kill the device over time. Luna: And helium is the tracer gas of choice because its atoms are tiny — second smallest after hydrogen. Lucas: Exactly. Helium molecules are about 0.26 nanometers in diameter. So if there's a path for helium to get through, water vapor — which is roughly 0.27 nanometers — will follow eventually. That's why it's so sensitive. Luna: So how does the actual test work? I imagine it's not just spraying helium at the device. Lucas: You've got two main methods. The more common one for high-reliability parts goes like this: you place the device inside a vacuum chamber, pump the air out, then spray helium outside the device. If there's a leak, helium gets sucked in through the opening, and a mass spectrometer inside the chamber detects it. Luna: And the other method is the opposite — you fill the device with helium first, then put it in a vacuum and measure what comes out. Lucas: Exactly. That's called the bomb method — you pressurize the part in a helium atmosphere for a few hours so helium diffuses into any internal cavities, then you transfer it to the vacuum chamber and measure the outgassing rate. Luna: And what's the pass/fail threshold most startups target? I've seen numbers like 1 times 10 to the minus 9 standard cubic centimeters per second. Lucas: That's the gold standard for hermetic — one times ten to the minus nine. For less critical stuff, like a consumer smartwatch that's just splash resistant, you might relax it to one times ten to the minus seven. But if you're building something that goes inside the human body, you want that nine. Luna: I worked with a startup last year making wearable insulin pumps — they had a gasket design that passed bubble testing but failed helium leak testing on about 8 percent of units. They traced it to a micro-molding defect in the silicone gasket. Lucas: That's a perfect example. Bubble testing — where you pressurize the device and submerge it in water looking for bubbles — can only catch leaks down to about ten to the minus five. Helium testing is a hundred thousand times more sensitive. Luna: And the cost? I remember the insulin pump startup spent about $18,000 on a benchtop helium leak detector from a company like Pfeiffer or Leybold. Lucas: That's typical. You're looking at $15,000 to $25,000 for the instrument, plus a vacuum pump and maybe a calibration leak standard. For a startup doing low-volume production, it might be cheaper to send parts to a test lab — about $50 to $150 per part for a full test. Luna: But if you're scaling to thousands of units, in-house makes sense pretty quickly. Lucas: Right. The per-unit cost drops fast once you've got the equipment. There's also a gotcha: residual helium. If you test a part and it fails, you need to let it outgas in a clean environment for a while before retesting. Otherwise you'll get a false positive — the mass spec picks up leftover helium from the previous test. Luna: That happened to that insulin pump startup — they thought they had a 30 percent failure rate until they realized the parts weren't fully degassed. Lucas: Yeah, operator training is real. You also need a calibrated leak standard — a tiny controlled leak that you run every day to verify the instrument is working. Without that, your numbers drift. Luna: So for a hardware startup, when should they first think about helium leak testing? Lucas: As soon as you have a prototype enclosure that's supposed to be sealed. Ideally before you commit to tooling for the final gasket or weld. Because if you find a design issue at the prototype stage, it's a CAD change. If you find it after you've ordered a thousand injection-molded housings, it's a write-off. Luna: And the alternative — pressure decay testing — is cheaper but less sensitive. You pressurize the part and monitor if pressure drops over time. That catches big leaks, but not the kind that kill a product after six months in a humid environment. Lucas: Exactly. Pressure decay is fine for a coffee maker that just needs to not leak water. But for anything with electronics that needs to survive rain, sweat, or sterilization, helium testing is the standard. Luna: What about optical leak detection? I've seen some systems use laser absorption to sniff out tracer gases. Lucas: That's emerging, but it's more expensive and less common at the startup level. Helium mass spec is still the workhorse. It's been around for decades, the equipment is mature, and the physics is well understood. Luna: One thing I want to circle back to — when we talk about 'hermetic' in the context of a startup, are we talking about fully sealed devices or ones with vents? Lucas: Great point. Many devices have vents with a membrane — like a Gore-Tex vent — that equalizes pressure while blocking water. Those are not hermetic. They have a controlled leak path. So helium testing would show a leak, and that's fine — you just need to know what your spec is. Hermetic means zero intentional leak path. Luna: So you'd test those vented enclosures differently — maybe just a spray test or IPX rating test. Lucas: Right. And that's where understanding your product's real environment matters. A sensor that sits on a bridge in Arizona probably doesn't need a hermetic seal. But the same sensor on a ship at sea? Totally different. Luna: I want to mention one more thing — the cost of not testing. A medical device startup I know skipped helium testing to save time. They shipped five hundred units. After a year, about sixty failed due to moisture ingress. The recall cost them over two hundred thousand dollars. Lucas: And that's not counting reputation damage. In hardware, reliability is the brand. A few thousand dollars on testing upfront is trivial compared to that. Luna: Speaking of keeping things going — a handful of listeners chip in monthly through buy me a coffee dot com slash fexingo, and that's literally what funds making this many of these episodes. No ads, just a small community. Lucas: Yeah, it's a quiet thing, but it means we can keep digging into topics like this without worrying about sponsors. Appreciate everyone who's part of that. Luna: Alright — back to the testing. So if a startup is ready to buy a helium leak detector, what should they look for in a system? Lucas: First, sensitivity range — make sure it goes down to ten to the minus nine. Second, cycle time — some systems take two minutes per part, others thirty seconds. For production, you want fast. Third, consider the size of the vacuum chamber — it needs to fit your largest enclosure with room for fixturing. Luna: And calibration? I've heard you need a leak standard, but some systems have built-in calibration. Lucas: Built-in calibration is nice but expensive. Most startups buy a separate calibrated leak — a little device with a known leak rate — and run it once a day. That's about $500 to $1,000. Luna: So between the detector, vacuum pump, fixturing, and calibration, you're looking at maybe $25,000 total. Lucas: Ballpark. And if you're doing a thousand units a year, that's $25 per unit in test equipment amortization. Compare that to a recall — it's a no-brainer. Luna: I think the takeaway is: helium leak testing is one of those things that seems like overkill until you need it. And once you need it, you wish you'd done it earlier. Lucas: Exactly. Test early, test often. The physics doesn't lie.